Articleshttps://doi.org/10.1038/s41477-019-0549-y

1PLECO (Plants and Ecosystems), Department of Biology, University of Antwerp, Antwerp, Belgium. 2Global Ecology Unit, CREAF–CSIC–UAB, Barcelona, Spain. 3US Geological Survey, Fort Collins Science Center, Fort Collins, CO, USA. 4CREAF, Barcelona, Spain. 5Department of Biological and Environmental Sciences, University of Gothenburg, Gothenburg, Sweden. 6Gothenburg Global Biodiversity Centre, Gothenburg, Sweden. 7Cornell Lab of Ornithology, Cornell University, Ithaca, NY, USA. 8Department of Biological Sciences, DePaul University, Chicago, IL, USA. 9Department of Systematic Zoology, Faculty of Biology, Adam Mickiewicz University in Poznań, Poznań, Poland. 10Institute for Agriculture and Forestry Systems in the Mediterranean, National Research Council of Italy (CNR-ISAFOM), Rende, Italy. 11Department of Innovation in Biological, Agro-food and Forest Systems, University of Tuscia, Viterbo, Italy. 12Department of Geography and Planning, School of Environmental Sciences, University of Liverpool, Liverpool, UK. 13DISAA, Università di Milano, Milan, Italy. *e-mail: marcos.fernandez-martinez@uantwerpen.be

Mast seeding—often called masting—has long intrigued biologists as one of the most bizarre reproductive behav-iours found in nature1,2. This behaviour consists of the

synchronous production of highly variable seed crops over time3. Masting has often been considered an evolutionary paradox because organisms that skip reproductive attempts should have lower fitness than those that reproduce at every opportunity4. Nonetheless, the fact that this reproductive behaviour is found in different lineages suggests that masting behaviour should be beneficial, at least under certain scenarios.

The most widely accepted hypotheses explaining the selective advantages of masting are all related to economies of scale5,6. Briefly, these hypotheses state that, in terms of fitness, it is more efficient for plants to produce a large number of seeds every few to several years than to produce a constant number every year. This general mechanism includes the predator satiation hypothesis2,7–9, where predators are starved during years of null or low reproduction and satiated during high reproduction mast years, leaving large num-bers of seeds intact. Another example is the pollination efficiency hypothesis5,10,11, which states that, particularly for wind-pollinated plants, saturating the atmosphere with pollen in a given year is more efficient than producing regular amounts of pollen each year to ensure pollination. Given that masting is present in only a mod-est percentage of plant species12, such economies of scale are appar-ently advantageous only under certain circumstances. What those circumstances are remains, so far, under debate.

The environmental stress hypothesis13 suggests that masting behaviour should be stronger under unfavourable growing condi-tions or limitation of resources—conditions under which econo-mies of scale should be more beneficial3,11,14. This is because plants

growing in unfavourable environments presumably experience more difficulties in acquiring the required resources to reproduce, as suggested by the resource accumulation hypothesis15,16. According to this hypothesis, plants growing under favourable conditions will be able to accumulate the required amount of resources every year and, therefore, present a regular pattern in seed production, without exhibiting any underlying negative temporal autocorrelation that could indicate resource depletion after reproduction15. The opposite is true for plants growing in unfavourable conditions, which will exhibit high interannual variability and negative temporal autocor-relation in seed production due to potential resource depletion after seeding. However, there is no current empirical evidence suggesting that species with higher interannual variability in fruit production are more likely to exhibit negative temporal autocorrelation than species that produce seeds more regularly. In contrast, weather variability has been found to be a key factor driving interan-nual variability in fruit production in many plant species11,17–20. Therefore, temporal patterns in weather events (that is, temporal variability and autocorrelation) could potentially shape the temporal patterns of fruit production21.

Foliar nutrient concentrations play a key role in plant ecophysi-ology and ecosystem functioning: photosynthetic rates are linked to foliar nitrogen (N) and phosphorus (P) concentrations22–24. Together with carbon, they are the basis of ecological stoichiom-etry25,26 and are fundamental parts of the elementome or the bio-geochemical niche27, useful for inferring ecological traits from the elemental composition of organisms28. N and P, as well as carbon (C), have been suggested to be potential resources determining seed production and masting behaviour14,29–31, because seeds and fruits are enriched with N and P compared with vegetative tissues32.

Nutrient scarcity as a selective pressure for mast seedingM. Fernández-Martínez   1,2*, I. Pearse3, J. Sardans2,4, F. Sayol   5,6, W. D. Koenig7, J. M. LaMontagne   8, M. Bogdziewicz   9, A. Collalti   10,11, A. Hacket-Pain   12, G. Vacchiano   13, J. M. Espelta4, J. Peñuelas   2,4 and I. A. Janssens   1

Mast seeding is one of the most intriguing reproductive traits in nature. Despite its potential drawbacks in terms of fitness, the widespread existence of this phenomenon suggests that it should have evolutionary advantages under certain circumstances. Using a global dataset of seed production time series for 219 plant species from all of the continents, we tested whether mast-ing behaviour appears predominantly in species with low foliar nitrogen and phosphorus concentrations when controlling for local climate and productivity. Here, we show that masting intensity is higher in species with low foliar N and P concentrations, and especially in those with imbalanced N/P ratios, and that the evolutionary history of masting behaviour has been linked to that of nutrient economy. Our results support the hypothesis that masting is stronger in species growing under limiting condi-tions and suggest that this reproductive behaviour might have evolved as an adaptation to nutrient limitations and imbalances.

NAturE PLANtS | VOL 5 | DECEMBER 2019 | 1222–1228 | www.nature.com/natureplants1222

ArticlesNaTUrE PlaNTs

Low foliar concentrations of N and P would imply lower photosyn-thetic rates22 that would in turn result in reduced accumulation of C, in addition to low accumulation of N and P. Thus, based on the resource budget model, plant species with low foliar N and P con-centrations would be expected to exhibit stronger masting behav-iour in terms of variability and negative temporal autocorrelation.

Foliar N and P have both been reported to be positively corre-lated with investment in plant reproduction33–35. Additionally, the interaction between foliar N and P (that is, the N/P ratio) has been suggested to be an important indicator of plant nutritional status, to promote vegetative productivity36–39 and to relate to reproduc-tive effort35 and ecological strategies27. The lower N/P ratio of seeds compared with vegetative tissues suggests that nutrient imbalances may constrain reproduction more than growth37. Previous studies have reported that plants and shoots producing flowers often exhibit lower N/P ratios than those that do not37,40, highlighting the role of nutrient imbalances in plant reproduction. However, no study has yet been focused on the evolutionary linkage between plant repro-duction and foliar nutrient stoichiometry over a large set of species.

Here, we hypothesize that mast seeding behaviour evolved pre-dominantly in plant species with low foliar N and P concentra-tions and high N/P ratios. To test this hypothesis, we first explored the relationship between temporal variability and autocorrelation in reproductive effort using a global dataset of 219 plant species from all of the continents, to investigate whether higher variability in seed production is related to higher potential resource depletion (negative temporal autocorrelation). We then fitted phylogenetic

linear models to test whether masting intensity—here defined as the combination of temporal variability and autocorrelation using the consecutive disparity (D) index41,42—was related to foliar N, P and N/P ratio, while controlling for the evolutionary history of the spe-cies and other potentially influencing factors such as local climate and the productivity of the regions where species were sampled. Finally, we tested the model of evolution of masting behaviour by means of ancestral character reconstructions, and explored whether it has evolved in concert with foliar N and P economies.

resultsTemporal variability, potential resource depletion and their evolutionary relationships. Our results indicated that temporal variability (here, calculated as the proportional variability index43 (PV); see Methods) in seed production is evolutionarily conserved (Extended Data Fig. 1a), showing a strong phylogenetic signal44. In contrast, lag 1 temporal autocorrelation (AR1, indicating potential resource depletion when it takes negative values) is not phyloge-netically conserved. Additionally, temporal variability and autocor-relation are not evolutionarily correlated. Hence, species exhibiting higher temporal variability do not necessarily exhibit any particu-lar pattern of temporal autocorrelation (Extended Data Fig. 1b). Masting intensity, defined as D in this study (see Methods), accounted for both features of masting behaviour (Fig. 1), tem-poral variability and potential resource depletion (negative AR1 coefficients), hence defining masting behaviour more broadly than the coefficient of variation (CV) alone, as it is typically assessed41.

–0.5 0 0.5

AR1

0.2

–2

–1

0

1

2

–2

–1

0

1

2

0.4 0.6 0.8 1.0

PVln

[D]

a b

c

ln[D

]

Masting intensity (D )Phylogenetic signal(λ = 0.44; P < 0.001) Evolutionary correlation

R = –0.13; P = 0.039

Evolutionary correlationR = 0.86; P < 0.001

0.1 >3.6

Encephalartos altensteiniiEncephalartos lehmanniiEncephalartos longifoliusEncephalartos villosusTsuga mertensianaTsuga heterophyllaAbies lasiocarpaAbies proceraAbies amabilisAbies magnifica

Abies grandis Abies concolor

Pseudotsuga menziesii

Larix occidentalis

Picea sitchensis

Picea glauca

Picea engelmannii

Picea mariana

Picea abies

Pinus edulis

Pinus albica

ulis

Pinus lambertia

na

Pinus montic

ola

Pinus banks

iana

Pinus pinea

Pinus r

esino

sa

Pinus s

ylves

tris

Pinus d

ensif

lora

Pinus c

onto

rta

Pinus

pon

dero

sa

Pinus

taed

a

Pinu

s pa

lust

ris

Cry

ptom

eria

japo

nica

Fitz

roya

cup

ress

oide

s

Thuj

a pl

icat

a

Cha

mae

cypa

ris o

btus

a

Juni

peru

s co

mm

unis

Juni

peru

s ox

yced

rus

Juni

peru

s sa

bina

Juni

peru

s ph

oeni

cea

Ara

ucar

ia a

ngus

tifol

ia

Ara

ucar

ia a

rauc

ana

Pru

mno

pity

s fe

rrug

inea

Sax

egot

haea

con

spic

ua

Dac

rydi

um c

upre

ssin

um

Pod

ocar

pus

tota

ra

Sm

ilax

aspe

ra

Pho

rmiu

m te

nax

Ast

roca

ryum

mex

ican

um

Oen

ocar

pus

map

ora

Rho

palo

styl

is s

apid

a

Chi

onoc

hloa

con

spic

ua

Chi

onoc

hloa

rub

raC

hion

ochl

oa c

rass

iusc

ula

Chionochloa teretifoliaC

hionochloaspiralisC

hionochloa rigida

Chionochloa pallens

Chionochloa oreophila

Chionochloa flavescens

Chionochloa m

acra

Drim

ys winteri

Laureliopsis philippiana

Beilschm

iedia pendula

Virola surinam

ensis

Virola sebifera

Desm

opsis panamensis

Annona spraguei

Guatteria dum

etorum

Gevuina avellana

Cissus erosa

Vitis rotundifolia

Terminalia oblonga

Mouriri m

yrtilloides

Amom

yrtus luma

Amom

yrtus meli

Eucalyptus regnans

Eucalyptus obliqua

Daphne gnidium

Dirca palustris

Luehea seemannii

Hampea appendiculata

Trattinnickia aspera

Tetragastris panamensis

Protium tenuifolium

Spondias radlkoferi

Spondias mombin

Anacardium excelsum

Astronium graveolens

Pistacia lentiscus

Pistacia terebinthus

Aesculus turbinata

Acer palmatum

Acer platanoides

Acer rufinerve

Acer saccharum

Swietenia macrophylla

Carapa guianensis Trichilia pallida

Trichilia tuberculata Guarea guidonia

Simarouba amaraPhellodendron amurense Zanthoxylum ekmanii Hippocratea volubilisPrionostemma asperaEucryphia cordifolia

Hirtella triandraRinorea sylvatica

Hybanthus prunifoliusCasearia arborea

Hasseltia floribunda

Salix myrsinifolia

Tetrapterys discolor

Nothofagus menziesii

Nothofagus cunninghamii

Nothofagus fusca

Nothofagus solandri

Nothofagus antarctica

Nothofagus nitida

Fagus grandifolia

Fagus sylvatica

Fagus crenata

Castanopsis cuspidata

Notholithocarpus densiflorus

Quercus ilex

Quercus suber

Quercus chrysolepis

Quercus laevis

Quercus m

yrtifo

lia

Quercus w

islize

ni

Quercus p

alustris

Quercus a

grifolia

Quercus k

elloggii

Quercu

s velu

tina

Quercu

s rub

ra

Querc

us p

etra

ea

Querc

us e

ngelm

annii

Que

rcus

gar

ryan

a

Que

rcus

den

tata

Que

rcus

mon

golic

a

Que

rcus

pub

esce

ns

Que

rcus

can

arie

nsis

Que

rcus

gem

inat

a

Que

rcus

mon

tana

Que

rcus

ste

llata

Que

rcus

loba

ta

Que

rcus

turb

inel

la

Que

rcus

alb

a

Que

rcus

cha

pman

iie

Mor

ella

cer

ifera

C

arya

ova

taC

arya

gla

bra

Jugl

ans

man

dshu

rica

Aln

us h

irsut

aA

lnus

inca

na

Car

pinu

s ja

poni

ca

Car

pinu

s co

rdat

aC

arpi

nus

laxi

flora

Car

pinu

s ts

chon

oski

iC

arpi

nus

betu

lus

Bet

ula

alle

ghan

iens

is

Bet

ula

max

imow

iczi

ana

Bet

ula

papy

rifer

a

Betula neoalaskana

Betula pendula

Betula platyphylla

Sorocea affinis

Brosim

um alicastrum

Poulsenia arm

ataR

ubus chamaem

orus R

ubus ulmifolius

Rubus cuneifolius

Rosa canina

Rosa m

ultifloraP

runus serotinaP

runus angustifolia

Sorbus am

ericana

Sorbus com

mixta

Sorbus aucuparia

Microberlinia bisulcata

Prioria copaifera

Hym

enaea courbaril

Acacia victoriae

Acacia tetragonophylla

Acacia aneura

Platymiscium

pinnatum

Platypodium elegans

Davilla nitida

Doliocarpus multiflorus

Heisteria concinna

Triplaris cumingiana

Aextoxicon punctatum

Cornus mas

Cornus sanguinea

Cornus amomum

Pouteria reticulata

Chrysophyllum cainito

Bertholletia excelsa

Arbutus unedo

Vaccinium myrtillus

Ilex aquifolium

Viburnum tinus

Schefflera morototoni

Dendropanax arboreus

Cordia alliodora

Cordia bicolor

Cordia lasiocalyx

Tabernaemontana arborea

Faramea occidentalis

Psychotria furcata

Psychotria horizontalis

Alseis blackiana

Randia armata

Phillyrea latifolia

Fraxinus mandshurica

Fraxinus excelsior

Jacaranda copaiaTabebuia rosea

Masting intensity (D)

Spe

cies

0 2 4 6 80

20

40

60

80

Fig. 1 | Masting behaviour intensity per species and its relationship with potential resource depletion (negative Ar1) and temporal variability (PV) of reproductive effort. a, Reconstructed evolution of masting intensity based on the disparity (D) index (using continuous character mapping; see Methods) and 219 species (n = 219 biologically independent samples). Inset: distribution of values of masting intensity. Phylogenetic signal was estimated using Pagel’s lambda (λ). b,c, Phylomorphospace plots showing the evolutionary Pearson’s correlation for masting intensity (natural logarithm of disparity (ln[D])) versus potential resource depletion (AR1; negative values indicate that resource depletion may happen) (b) and temporal variability (PV) (c). In b and c, t-values of the correlations were −2.08 and 25.19 (217 degrees of freedom (DF)), respectively. The plant silhouettes in a were drawn by F.S. and are available at PhyloPic (http://phylopic.org).

NAturE PLANtS | VOL 5 | DECEMBER 2019 | 1222–1228 | www.nature.com/natureplants 1223

Articles NaTUrE PlaNTs

Masting intensity was also preserved phylogenetically (Fig. 1). Our results suggest that most species exhibit low or intermediate masting intensity, while only a few exhibit strong masting behaviour (see histogram in Fig. 1a).

Estimating controls and the mode of evolution of masting behav-iour. Phylogenetic models indicated a statistically significant nega-tive interaction between foliar N and P, explaining variability in masting intensity across species (Table 1 and Fig. 2). Model results indicated that when foliar P is low, increasing foliar N increases masting intensity as N/P increases. The inverse situation (high masting intensity at very low N/P ratios) is also possible accord-ing to model results, despite being a less likely scenario (we only had 19 species with N/P < 8 in our dataset). The highest masting intensity was found in species with high N/P, while the lowest was found in species with high foliar N and P concentrations. Species with low foliar N and P concentrations showed intermediate val-ues of masting intensity. Temporal variability of the local climate or productivity of the region from which species were sampled did not explain variation in masting intensity. However, mean annual precipitation (MAP) was positively related to masting intensity (Table 1). Temporal variability in seed production was negatively related to mean annual temperature (MAT) and positively related to temporal variability of annual precipitation. Additionally, spe-cies with higher foliar P exhibited lower temporal variability. When assessing temporal variability using the CV, we found that species from regions with higher temporal variability in productivity also presented a higher CV of seed production (standardized model coefficient, β = 0.22 ± 0.08; t = 2.96; P = 0.004; phylogenetic signal, λ = 0.58; n = 168 species). Potential resource depletion was more likely to occur in species living in climates exhibiting negative tem-poral autocorrelation in annual precipitation (Table 1). Interestingly, the model also showed a positive interaction between foliar N and P, indicating that potential resource depletion after seeding was more likely to occur in species with low foliar N and P.

Our analyses also revealed the evolutionary processes linking masting behaviour and foliar nutrient concentrations (Extended Data Fig. 2). We found that adaptive processes—Ornstein–Uhlenbeck models based on ancestral character reconstructions (Extended Data Fig. 3)—are the models that better explain cur-rent patterns in masting intensity linked to foliar nutrient con-centrations. Adaptive Ornstein–Uhlenbeck models explaining

the evolutionary history of masting intensity with N, P and N × P (a measure of the total N and P foliar concentration) assume that species evolved towards different optimum values depending on whether they present high or low masting intensity (masting and non-masting species) (OUM model). However, a single optimum for all species was equally likely for foliar N, P and N × P (<2 units of second-order Akaike information criterion (AICc) between models were found (OU1); see Methods). Nonetheless, given that OUM models with different optimum values performed equally well for foliar N, P and N × P (Extended Data Fig. 2), OUM models are used to illustrate our results. The model that best fitted the data for foliar N/P was an adaptive Ornstein–Uhlenbeck model with dif-ferent optimum and phenotypic variation for high and low masting intensity species (the OUMV model).

Results from Ornstein–Uhlenbeck models indicated that mast-ing species had, on average, 9.5 ± 0.4 and 18.3 ± 0.5% lower foliar N and P concentrations, respectively, than non-masting species (Fig. 3 and Extended Data Fig. 4). The foliar N/P ratio was 11.4 ± 0.2% (mean ± s.e.m.) higher in masting species, while N × P (the com-bined availability of N and P) was 28.7 ± 1.0% lower than in non-masting species. These results were consistent when using five different thresholds of masting intensity (see Methods) to classify species as masting or non-masting species (Extended Data Fig. 4). Like masting intensity, foliar N and P concentrations were both pre-served throughout the phylogeny (Extended Data Fig. 5).

DiscussionRole of foliar nutrients in seed production. Our results suggest that masting intensity co-evolved with species-specific optimal foliar N and P concentrations and that species with lower N (and especially lower P), and therefore high N/P, exhibit higher mast-ing intensity (Fig. 3). Hence, our analyses supported our initial hypothesis stating that masting behaviour evolved predominantly

Table 1 | Model summary of the phylogenetic linear models based on 168 species

β s.e.m. t P λ

Masting intensity (D)

MAP 0.22 0.09 2.44 0.016

N 0.04 0.10 0.38 0.708

P −0.04 0.10 −0.42 0.674

N × P −0.19 0.07 −2.76 0.007 0.47

temporal variability (PV)

MAPPV 0.16 0.08 2.05 0.042

MAt −0.18 0.09 −2.07 0.040

P −0.16 0.08 −2.03 0.044 0.41

Potential resource depletion (Ar1)

MAPAr1 0.27 0.08 3.42 0.001

N −0.09 0.10 −0.85 0.001

P −0.05 0.10 −0.45 0.008

N × P 0.22 0.07 3.19 0.002 0

Foliar N (percentage of dry weight)

1.0

1.5

2.0

2.5

3.0

Foliar P (percentage of dry weight)

0.1

0.2

0.3

Mas

ting

inte

nsity

(D

)

0.5

1.0

1.5

2.0

2.5

Fig. 2 | 3D graph showing the interaction between foliar N and P and masting intensity. Combined high foliar N and P concentrations decreased the masting intensity, although the highest masting intensity was found in species with the highest N/P imbalances. The response surface of the negative interaction between N and P (see Table 1 for statistics) was estimated from a phylogenetic linear model based on 168 species (n = 168 biologically independent samples). See Methods for further information on the models.

NAturE PLANtS | VOL 5 | DECEMBER 2019 | 1222–1228 | www.nature.com/natureplants1224

ArticlesNaTUrE PlaNTs

in plant species with low foliar N and P concentrations and high N/P ratios. This observed evolutionary pattern may have originated because of different underlying mechanisms driven by environ-mental and physiological constraints. One potential mechanism explaining these findings could be the physiological role of foliar N and P concentrations in plants. Foliar N is well known for being the primary limiting nutrient for reproduction33,34 and vegetative growth45,46. Elevated foliar P concentrations and low N/P ratios have been shown to allow larger seed crops in multiple species14,35,40,47, coinciding with fast-growing species according to the growth rate hypothesis25. P is also essential to maintain water-use efficiency and growth, particularly during drought conditions48. Both nutrients are essential elements of ribosomes and therefore play a major role in organismal metabolism49. The production of seeds, which are struc-tures enriched with N and P32, may potentially benefit from high concentrations of nutrients even more than vegetative tissues.

Low foliar N and P concentrations and high N/P imbalance are often indicative of unfavourable environments for plant growth14,50,51, such as infertile soils. These poor growing conditions are often related to dry or cold climates where decomposition of organic mat-ter is constrained, rates of nutrient mineralization and weathering are low48,52, or disturbances (for example, wildfires) are frequent53. However, differences in foliar N and P concentrations may not

always reflect differences in nutrient availability across species, as coexisting species may have different elemental compositions (the biogeochemical niche hypothesis27). Nonetheless, plants with high foliar N and P concentrations must either be growing in environ-ments without nutrient limitations or have developed mechanisms that allow them to sustain high foliar nutrient concentrations even if they are growing under unfavourable conditions. In any case, these higher concentrations of nutrients should confer a competitive advantage in terms of C acquisition, because higher N and P con-centrations are, on average, linked to higher photosynthetic rates22.

Evolutionary history of masting intensity and foliar nutrients. Being capable of investing more in reproduction does not explain masting behaviour by itself, because equal average seed crops over time could lead to different reproductive behaviours14. The neces-sary link between the ecological stoichiometry and masting theoret-ical backgrounds lies in the resource accumulation hypothesis15,16. Plant species with lower or imbalanced availability of N and P may present more difficulties in acquiring the necessary amounts of C, N and P to successfully produce seeds regularly, thereby mechanisti-cally producing a reproductive behaviour aligned with high masting intensity: high interannual variability and negative temporal auto-correlation (that is, potential resource depletion). The combination

0.150

0.175

0.200

0.225

1.50 1.75 2.00 2.25

Foliar N (percentage of dry weight)

a

bP < 0.001 P < 0.001

P < 0.001 P < 0.001

Fol

iar

P (

perc

enta

ge o

f dry

wei

ght)

0.2

0.3

0.4

0.6

0.5

10 11 12 13

Foliar N/P

Fol

iar

N ×

P

1.5

1.8

2.1

2.4

Fol

iar

N (

perc

enta

ge o

f dry

wei

ght)

0.15

0.20

0.25

Foliar P

(percentage of dry weight)

10

11

12

13

Fol

iar

N/P

0.2

0.3

0.4

0.5

0.6

Foliar N

× P

Non-mastingMasting

Fig. 3 | Different optimum values of foliar N and P for subsets of masting and non-masting species. a, Model results for foliar N and P concentrations for a subset of species identified as masting (high masting intensity) and non-masting based on the 20th and 80th percentiles of the D distribution (see Methods). b, Model results for foliar N/P and N × P (overall nutrient availability) for masting and non-masting species. In a and b, n = 1,000 and 927 independent stochastic character mapping simulations were used, respectively (see Extended Data Fig. 4). For foliar N/P, the results from OUMV models were used, while for N, P and N × P, OUM models were used (see Extended Data Figs. 2 and 4). Differences between groups were calculated based on two-sided paired t-tests (t-values were −24.34, −47.44 (both on 999 DF), 66.05 and −39.24 (both for 926 DF) for N, P, N/P and N × P, respectively). Dashed lines indicate the mean x and y values for the different groups. Violin plots represent density distributions of the data. Boxplots show the median (centre line), 25th and 75th quartiles (hinges), 1.5 times the interquartile range from the hinges (whiskers) and values outside 1.5 times the interquartile range (points). See Methods for further information.

NAturE PLANtS | VOL 5 | DECEMBER 2019 | 1222–1228 | www.nature.com/natureplants 1225

Articles NaTUrE PlaNTs

of low and imbalanced nutrient availability (causing high variabil-ity in seed production and potential resource depletion after crops) and environmental variability (which synchronizes the reproduc-tion of individuals through weather cues6,18) would finally shape the reproductive behaviour of masting species. In fact, our results also revealed that MAP, its temporal variability and its autocorrelation, are related to the reproductive behaviour of plant species (Table 1). These results indicate that even nutrient-rich species can present masting behaviour if they grow under climates with highly fluctuat-ing weather conditions. Therefore, we suggest that the interaction between weather conditions and the availability of nutrients (both of which condition photosynthetic rates) is the triggering factor of the common nonlinear (often exponential-like) response21 between seed production and weather variability in masting species7,17,20.

As a result of the environmental and physiological constraints, species may have been selected to exhibit distinctive reproductive behaviours in order to increase their fitness. Nutrient-rich species may not have developed a resource-conservative masting-like reproduction strategy because of their capacity to produce abun-dant seeds with regularity, avoiding losing reproduction attempts. Instead, because the investment in reproduction in terms of C and nutrients should be proportionally more expensive for nutrient-poor or nutrient-imbalanced species, the pressure to exhibit more cost-efficient reproduction may have selected such species to produce fewer but larger reproductive events, to take advantage of one or more economies of scale5,6,10. As a side effect of these massive repro-ductive events, negative temporal autocorrelation would also have appeared in masting species because of potential resource depletion (Extended Data Fig. 1). Hence, these particular traits would have been preserved throughout evolutionary history (Fig. 1) because foliar functional traits and masting intensity have co-evolved (Fig. 3 and Extended Data Fig. 4). Nonetheless, our results do not discard other potential selective pressures that may have triggered the evolu-tion of masting behaviour. Some species may have been selected to mast to improve their pollination efficiency54, to escape seed preda-tion from voracious predators7,8 or because of interspecific competi-tion in different ways (for example, seedling establishment). Also, a selective pressure towards reproducing more constantly could occur in animal-pollinated species, where a more constant production of flowers would favour populations of pollinators and, hence, pollina-tion. Further research on the reproductive behaviours of early plants such as bryophytes55 and taxa from different realms, using concepts from the masting literature, may facilitate a better understanding of the evolution of different reproductive behaviours in nature.

MethodsDatasets. Masting database. Data on interannual reproductive effort (seed or fruit production) were compiled by searching the Web of Science, scanning literature citations in published papers to look for more records of reproductive effort, contacting managers of wildlife surveys, forestry districts and regional seed surveys, and soliciting datasets in the Ecolog listserv. See Pearse et al.13 for more information on the data collection methods and characteristics of the dataset.

We only included records when: (1) data were available for more than four consecutive years per species at a given site (with clear geographical coordinates); (2) records could be assigned clearly to plants of a particular species; (3) records were not measured in such an indirect way that their accuracy could be jeopardized (for example, anticipated correlations with gamete abundance); (4) data represented seed or fruit production, or inflorescence set only for those cases where inflorescences were strongly linked to seed or fruit production; and (5) records were from iteroparous perennial plants whose seed set could not be explained by changes in population size. We did not include records of pollen production, nor records from agricultural settings.

Overall, the reproductive effort dataset contained 1,084 records of reproductive effort, including 363 plant species (trees, shrubs, vines, grasses and herbs) from 205 studies, ranging from 1900–2014 and covering the six majorly vegetated continents (Africa (17), North America (466), Europe (280), Japan (68), New Zealand (67), Central America (118) and other regions (68)). On average, records were 11.9 years long, although 131 studies had more than 20 years of data13.

For each site and species, we calculated the proportional variability index in seed production (PV)43 as a measure of temporal variability. The PV index

overcomes several statistical and mathematical issues of the CV index43. The PV index was calculated, for each time series, as:

PV ¼ 2P

zn n� 1ð Þ

where z was calculated as:

z ¼ 1� min zi; zj� �

max zi; zj� �

Where z represents the list of individual values (seed production per year), from each time series, from which to calculate the pairwise comparisons, and n indicates the number of values in a time series. We also calculated the lag 1 temporal autocorrelation (AR1) as a measure of potential resource depletion using the acf function in R, and the consecutive disparity index (D)41,42 as a measure of masting intensity. We defined masting intensity as D because, like the PV index, this index also overcomes several statistical and mathematical limitations of the coefficient of variation (CV = standard deviation × mean−1, commonly used to describe masting12,13), including dependence on the mean, dependence on the length of the time series and bias associated with non-Gaussian data. More importantly, we used D because it combines two of the main features describing masting behaviour: temporal variability and lag 1 autocorrelation (AR1) in seed production41. D was defined as:

D ¼ 1n� 1

Xn�1

i¼1

lnpiþ1 þ kpi þ k

Where pi is the series value (seed production in our case) at time i, n is the series length and k is a constant (often one unit) to avoid numerical indetermination in time series with 0 values. D is high when temporal variability is high and lag 1 temporal autocorrelation is negative (that is, strong masting behaviour, showing potential resource depletion after large seed crops19,30). Conversely, D is low when temporal variability is low and temporal autocorrelation is close to zero or positive (that is, describing a constant pattern of seed production and, hence, no masting behaviour). Given that the CV is still the most widely used index for assessing temporal variability in masting studies, we also calculated the CV of seed production per species and site to compare its results with those shown by the PV index. When multiple records from the same species were available, we calculated the average masting intensity (D), temporal variability (PV and CV) and potential resource depletion coefficient (AR1) per species.

Climate, productivity and foliar nutrients. We extracted MAT (°C) and MAP (mm yr−1) and their seasonality (MATs (standard deviation of monthly values × 100) and MAPs (CV)) for each location in our masting database from the WorldClim 2 database56. The climate database contains long-term means (1950–2000), calculated on a 30 arcsec grid. Data (for variability) and autocorrelation (for temperature and precipitation) were extracted from the CRU TS version 3.25 dataset57. To estimate site productivity, we used a remotely sensed gross primary production (GPP) database58. For annual temperature and precipitation of the CRU TS and GPP global databases, we calculated D, PV and AR1 indices for each pixel. We then extracted climate (that is, MATD, MAPD, MATPV, MAPPV, MATAR1 and MAPAR1) and productivity (that is, GPPD, GPPPV and GPPAR1) data for each site and species in our masting database and calculated the average per species (Extended Data Fig. 6). We used these variables in our statistical analyses to control for site-specific differences in temporal variability, as well as autocorrelation patterns of climate and productivity.

Data on the foliar concentration of N and P for the species in our database were gathered from the TRY Plant Trait Database (https://www.try-db.org/TryWeb/Home.php)59, BIEN database (http://bien.nceas.ucsb.edu/bien/)60, ICP Forests database on foliar elemental concentration61 (www.icp-forests.net) and Catalan Forest Inventory (http://www.creaf.uab.cat/iefc/). To estimate an average value per species, we first calculated the mean value per species and database. We then merged all databases and calculated the mean values per species. Species names in our database were checked and corrected using The Plant List database in the R package Taxonstand62. Phylogenetic analyses were performed using the plant phylogeny provided by Qian and Jin63. Of the 363 species in the masting database, 219 species names (~60%) matched those in the phylogenetic tree; therefore, only these were used for further analyses. Analyses using foliar N and P data were restricted to the 168 species (~46%) for which we could find data63.

Data analyses. Evolutionary link between masting intensity, temporal variability and autocorrelation. To explore how well the phylogenetic ancestry can explain masting behaviour, we first estimated the phylogenetic signal (that is, the tendency for related species to resemble each other more than they resemble species drawn at random from the phylogenetic tree) of masting intensity (D), PV and AR1 using the phylosig function in the R64 package phytools65. The phylogenetic signal was assessed by the λ metric, which varies from 0 (where phylogenetic and trait similarity are totally independent) to 1 (where the traits are completely explained by shared ancestry). We then used continuous mapped

NAturE PLANtS | VOL 5 | DECEMBER 2019 | 1222–1228 | www.nature.com/natureplants1226

ArticlesNaTUrE PlaNTs

phylogenetic trees (contMap function in the R package phytools) to visualize their phylogenetic signal. Finally, we explored the evolutionary relationship between temporal variability, temporal autocorrelation and masting intensity using pairwise correlations correcting for the phylogeny. Relationships were shown using phylomorphospace plots65, which depict each species as a data point in a trait space, together with the phylogenetic relationship of each species point.

Controls of masting intensity and its mode of evolution. We first tested whether masting intensity (D) was related to climate, productivity, foliar N and P concentrations and their interaction. To do so, we fitted phylogenetic linear models in which the response variable was masting intensity (D) and the predictors were: foliar N and P concentrations and their interaction; MAT, MAP, MATs, MAPs, MATD and MAPD for climate; and GPP and GPPD for productivity. Phylogenetic models were fitted optimizing λ (that is, the strength of phylogenetic signal) and using the phylolm function in the R phylolm package66. The final model was achieved by removing the least significant terms from the full model, in a step-by-step process, until all variable estimates were significant. The same models were fitted for the PV and AR1, but changing the predictors from D to PV or AR1, respectively (for example, MATPV instead of MATD when predicting PV). Because CV has been widely used to assess temporal variability, we also fitted a model using CV as the predictor variable to compare its results with those of the model using the PV index. Masting intensity and PV were transformed to natural logarithms to normalize model residuals. We used the package visreg67 to visualize the model results.

Evolution of masting intensity and foliar N and P and their interaction. To test the hypothesis that masting behaviour has evolved as an adaptation to nutrient imbalances and low foliar N and P concentrations, we performed a three-step analysis: (1) we classified species as masting and non-masting; (2) we reconstructed the ancestral state between the two types of reproducing behaviour; and (3) we fitted different evolutionary models to test whether foliar concentrations of N, P and N/P ratio and N × P (as a measure of the overall availability of nutrients) evolved under the reconstructed discrete selective regimes (masting or non-masting).

To define masting behaviour, and as a test for the robustness of our results, we classified species as subsets that represent masting (high temporal variability and strong potential resource depletion) or non-masting (low temporal variability and no resource depletion) behaviours based on five different thresholds of masting intensity (D), selecting only the lower and upper bounds for the analyses and discarding the intermediate species. The selected percentile thresholds were: (1) from 0–33% non-masting (n = 38 species) and from 66–100% masting (n = 58); (2) from 0–25% non-masting (n = 32) and from 75–100% masting (n = 43); (3) from 0–20% non-masting (n = 27) and from 80–100% masting (n = 34); (4) from 0–15% non-masting (n = 22) and from 85–100% masting (n = 26); and (5) from 0–10% non-masting (n = 16) and from 90–100% masting (n = 17).

To reconstruct ancestral states of masting behaviour, for each of the five classifications, we performed stochastic character mappings68, which reconstruct the state of the ancestors of a phylogeny based on the observed traits of the current species and the phylogenetic structure. Ancestral reconstructions were performed using the make.simmap function in the phytools R package65, simulating 1,000 stochastic ancestral reconstructions, specifying equal rates of transition among the character states and using the Markov chain Monte Carlo method.

Once we had the 1,000 stochastic character mappings for each masting classification, we performed generalized Ornstein–Uhlenbeck Hansen models to test whether the inferred evolutionary trajectories in foliar concentrations of N, P, N/P and N × P (hereafter ‘continuous traits’) were associated with the two alternative masting behaviour strategies (hereafter ‘ancestral states’) and whether they followed an adaptive (Ornstein–Uhlenbeck) or random (Brownian motion) model of evolution69–71. To do so, we used the Ornstein–Uhlenbeck function from the OUwie R package72. We fitted five different models using the 1,000 ancestral reconstructions mentioned above for each classification. The five different models represent different types of underlying evolutionary processes, being: (1) single-state Brownian motion models (BM1), where evolutionary rates for the continuous traits are equal for all ancestral states; (2) Brownian motion models with different evolutionary rates for each ancestral state (BMS); (3) Ornstein–Uhlenbeck models with a single optimal value for the continuous traits for all ancestral states (OU1); (4) Ornstein–Uhlenbeck models with different optimal values but a single α (the strength of the pull towards the optimal values of the trait) and single θ (the rate of phenotypic variation around the optimal value) for each state (OUM); and (5) Ornstein–Uhlenbeck models that assumed different optimal values with multiple rates of phenotypic evolution (θ) for each state (OUMV).

Models containing negative eigenvalues (non-sound models) were deleted when summarizing our results28. Different evolutionary models were compared using AICc among all sound models. Those models with the lowest AICc were considered to fit the data best. For models assuming different optimal values of foliar N, P, N/P and N × P for masting and non-masting species (that is, the OUM and OUMV models), we calculated the geometric mean of the percentage differences of each model. Statistical differences in optimal values estimated between subsets of masting and non-masting species by the Ornstein–Uhlenbeck models with different state means (OUM and OUMV) were tested using paired t-tests. Given that the results pointed out in the same direction (see Extended

Data Fig. 4) for all masting intensity thresholds, we only show those from the intermediate threshold (0–20% for non-masting and 80–100% for masting). These analyses used the 168 species for which we had data for masting intensity and foliar N and P concentrations. All statistical analyses were performed with R statistical software version 3.5.1 (ref. 64).

Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availabilityData supporting the findings of this study can be found at https://doi.org/10.6084/m9.figshare.9863006.v1 and https://www.sciencebase.gov/catalog/item/5d9f7f49e4b036616294495d. Any use of trade, firm or product names is for descriptive purposes only and does not imply endorsement by the US Government.

Received: 8 April 2019; Accepted: 8 October 2019; Published online: 2 December 2019

references 1. Salisbury, E. The Reproductive Capacity of Plants (Bell, 1942). 2. Silvertown, J. W. The evolutionary ecology of mast seeding in trees.

Biol. J. Linn. Soc. 14, 235–250 (1980). 3. Kelly, D. & Sork, V. L. Mast seeding in perennial plants: why, how, where?

Annu. Rev. Ecol. Syst. 33, 427–447 (2002). 4. Waller, D. M. Models of mast fruiting in trees. J. Theor. Biol. 80,

223–232 (1979). 5. Norton, D. A. & Kelly, D. Mast seeding over 33 years by Dacrydium

cupressinum Lamb. (rimu) (Podocarpaceae) in New Zealand: the importance of economies of scale. Funct. Ecol. 2, 399–408 (1988).

6. Pearse, I. S., Koenig, W. D. & Kelly, D. Mechanisms of mast seeding: resources, weather, cues, and selection. New Phytol. 212, 546–562 (2016).

7. Espelta, J. M., Cortés, P., Molowny-horas, R., Sánchez-Humanes, B. & Retana, J. Masting mediated by summer drought reduces acorn predation in Mediterranean oak forests. Ecology 89, 805–817 (2008).

8. Espelta, J. M. et al. Beyond predator satiation: masting but also the effects of rainfall stochasticity on weevils drive acorn predation. Ecosphere 8, e01836 (2017).

9. Donaldson, J. S. Mast-seeding in the cycad genus Encephalartos: a test of the predator satiation hypothesis. Oecologia 94, 262–271 (1993).

10. Smith, C. C., Hamrick, J. L. & Kramer, C. L. The advantage of mast years for wind pollination. Am. Nat. 136, 154–166 (1990).

11. Fernández-Martínez, M., Belmonte, J., Maria Espelta, J. & Espelta, J. M. Masting in oaks: disentangling the effect of flowering phenology, airborne pollen load and drought. Acta Oecologica 43, 51–59 (2012).

12. Herrera, C., Jordano, P., Guitián, J. & Traveset, A. Annual variability in seed production by woody plants and the masting concept: reassessment of principles and relationship to pollination and seed dispersal. Am. Nat. 152, 576–594 (1998).

13. Pearse, I. S., LaMontagne, J. M. & Koenig, W. D. Inter-annual variation in seed production has increased over time (1900–2014). Proc. R. Soc. B Biol. Sci. 284, 20171666 (2017).

14. Fernández-Martínez, M., Vicca, S., Janssens, I. A., Espelta, J. M. & Peñuelas, J. The role of nutrients, productivity and climate in determining tree fruit production in European forests. New Phytol. 213, 669–679 (2016).

15. Isagi, Y., Sugimura, K., Sumida, A. & Ito, H. How does masting happen and synchronize? J. Theor. Biol. 187, 231–239 (1997).

16. Ye, X. & Sakai, K. A new modified resource budget model for nonlinear dynamics in citrus production. Chaos Solitons Fract. 87, 51–60 (2016).

17. Pérez-Ramos, I. M., Ourcival, J. M., Limousin, J. M. & Rambal, S. Mast seeding under increasing drought: results from a long-term data set and from a rainfall exclusion experiment. Ecology 91, 3057–3068 (2010).

18. Fernández-Martínez, M., Vicca, S., Janssens, I. A., Espelta, J. M. & Peñuelas, J. The North Atlantic Oscillation synchronises fruit production in western European forests. Ecography 39, 864–874 (2016).

19. Sork, V. L., Bramble, J. & Sexton, O. Ecology of mast-fruiting in three species of North American deciduous oaks. Ecology 74, 528–541 (1993).

20. Perez-Ramos, I. M., Padilla-Díaz, C. M. & Koenig, W. D. & Marañón, T. Environmental drivers of mast-seeding in Mediterranean oak species: does leaf habit matter? J. Ecol. 103, 691–700 (2015).

21. Fernández-Martínez, M., Bogdziewicz, M., Espelta, J. M. & Peñuelas, J. Nature beyond linearity: meteorological variability and Jensen’s inequality can explain mast seeding behavior. Front. Ecol. Evol. 5, 134 (2017).

22. Wright, I. J. et al. The worldwide leaf economics spectrum. Nature 428, 821–827 (2004).

23. Domingues, T. F. et al. Co-limitation of photosynthetic capacity by nitrogen and phosphorus in West Africa woodlands. Plant Cell Environ. 33, 959–980 (2010).

NAturE PLANtS | VOL 5 | DECEMBER 2019 | 1222–1228 | www.nature.com/natureplants 1227

Articles NaTUrE PlaNTs

24. Field, C., Merino, J. & Mooney, H. A. Compromises between water-use efficiency and nitrogen-use efficiency in five species of California evergreens. Oecologia 60, 384–389 (1983).

25. Elser, J. J. et al. Nutritional constraints in terrestrial and freshwater food webs. Nature 408, 578–580 (2000).

26. Sardans, J. et al. Factors influencing the foliar elemental composition and stoichiometry in forest trees in Spain. Perspect. Plant Ecol. Evol. Syst. 18, 52–69 (2016).

27. Peñuelas, J. et al. The bioelements, the elementome, and the biogeochemical niche. Ecology 100, e02652 (2019).

28. Fernández-Martínez, M. et al. Nutrient-rich plants emit a less intense blend of volatile isoprenoids. New Phytol. 220, 773–784 (2018).

29. Camarero, J. J., Albuixech, J., López-Lozano, R., Casterad, M. A. & Montserrat-Martí, G. An increase in canopy cover leads to masting in Quercus ilex. Trees 24, 909–918 (2010).

30. Fernández-Martínez, M., Garbulsky, M., Peñuelas, J., Peguero, G. & Espelta, J. M. Temporal trends in the enhanced vegetation index and spring weather predict seed production in Mediterranean oaks. Plant Ecol. 216, 1061–1072 (2015).

31. Sala, A., Hopping, K., McIntire, E. J. B., Delzon, S. & Crone, E. E. Masting in whitebark pine (Pinus albicaulis) depletes stored nutrients. New Phytol. 196, 189–199 (2012).

32. Reekie, E. G. & Bazzaz, F. A. Reproductive effort in plants. 2. Does carbon reflect the allocation of other resources? Am. Nat. 129, 897–906 (1987).

33. Miyazaki, Y. et al. Nitrogen as a key regulator of flowering in Fagus crenata: understanding the physiological mechanism of masting by gene expression analysis. Ecol. Lett. 17, 1299–1309 (2014).

34. Han, Q., Kabeya, D., Iio, A., Inagaki, Y. & Kakubari, Y. Nitrogen storage dynamics are affected by masting events in Fagus crenata. Oecologia 174, 679–687 (2013).

35. Fujita, Y. et al. Low investment in sexual reproduction threatens plants adapted to phosphorus limitation. Nature 505, 82–86 (2014).

36. Sardans, J., Rivas-Ubach, A. & Peñuelas, J. The C:N:P stoichiometry of organisms and ecosystems in a changing world: a review and perspectives. Perspect. Plant Ecol. Evol. Syst. 14, 33–47 (2012).

37. Güsewell, S. N. P ratios in terrestrial plants: variation and functional significance. New Phytol. 164, 243–266 (2004).

38. McGroddy, M. E., Daufresne, T. & Hedin, O. L. Scaling of C:N:P stoichiometry in forests worldwide: implications of terrestrial redfield-type ratios. Ecology 85, 2390–2401 (2004).

39. Sardans, J. et al. Foliar and soil concentrations and stoichiometry of nitrogen and phosphorous across European Pinus sylvestris forests: relationships with climate, N deposition and tree growth. Funct. Ecol. 30, 676–689 (2016).

40. Eckstein, R. L. & Karlsson, P. S. Above-ground growth and nutrient use by plants in a subarctic environment: effects of habitat, life-form and species. Oikos 79, 311–324 (1997).

41. Fernández-Martínez, M., Vicca, S., Janssens, I. A., Martín-Vide, J. & Peñuelas, J. The consecutive disparity index, D, as measure of temporal variability in ecological studies. Ecosphere 9, e02527 (2018).

42. Martín-Vide, J. Notes per a la definició d’un índex de “desordre“ en pluviometria. Soc. Catalana Geogr. 7, 89–96 (1986).

43. Heath, J. P. Quantifying temporal variability in population abundances. Oikos 115, 573–581 (2006).

44. Losos, J. B. Phylogenetic niche conservatism, phylogenetic signal and the relationship between phylogenetic relatedness and ecological similarity among species. Ecol. Lett. 11, 995–1003 (2008).

45. Elser, J. J. et al. Global analysis of nitrogen and phosphorus limitation of primary producers in freshwater, marine and terrestrial ecosystems. Ecol. Lett. 10, 1135–1142 (2007).

46. LeBauer, D. S. & Treseder, K. K. Nitrogen limitation of net primary productivity in terrestrial ecosystems is globally distributed. Ecology 89, 371–379 (2008).

47. Erel, R. Flowering and fruit set of olive trees. Am. Soc. 133, 639–647 (2008). 48. Sardans, J. & Peñuelas, J. Drought changes phosphorus and potassium

accumulation patterns in an evergreen Mediterranean forest. Funct. Ecol. 21, 191–201 (2007).

49. Elser, J. J. et al. Growth rate–stoichiometry couplings in diverse biota. Ecol. Lett. 6, 936–943 (2003).

50. Fernández-Martínez, M. et al. Nutrient availability as the key regulator of global forest carbon balance. Nat. Clim. Change 4, 471–476 (2014).

51. Nogueira, C. et al. Leaf nutrients, not specific leaf area, are consistent indicators of elevated nutrient inputs. Nat. Ecol. Evol. 3, 400–406 (2019).

52. Smaill, S. J., Clinton, P. W., Allen, R. B. & Davis, M. R. Climate cues and resources interact to determine seed production by a masting species. J. Ecol. 99, 870–877 (2011).

53. Saura-Mas, S. & Lloret, F. Linking post-fire regenerative strategy and leaf nutrient content in Mediterranean woody plants. Perspect. Plant Ecol. Evol. Syst. 11, 219–229 (2009).

54. Kelly, D., Hart, D. & Allen, R. B. Evaluating the wind pollination benefits of mast seeding. Ecology 82, 117–126 (2001).

55. Fernández‐Martínez, M. et al. Towards a moss sclerophylly continuum: evolutionary history, water chemistry and climate control traits of hygrophytic mosses. Funct. Ecol. https://doi.org/10.1111/1365-2435.13443 (2019).

56. Fick, S. E. & Hijmans, R. J. WorldClim 2: new 1-km spatial resolution climate surfaces for global land areas. Int. J. Climatol. 37, 4302–4315 (2017).

57. Harris, I., Jones, P. D. D., Osborn, T. J. J. & Lister, D. H. H. Updated high-resolution grids of monthly climatic observations—the CRU TS3.10 dataset. Int. J. Climatol. 34, 623–642 (2013).

58. Zhang, Y. et al. A global moderate resolution dataset of gross primary production of vegetation for 2000–2016. Sci. Data 4, 170165 (2017).

59. Kattge, J. et al. TRY—a global database of plant traits. Glob. Change Biol. 17, 2905–2935 (2011).

60. Maitner, B. S. et al. The bien r package: a tool to access the Botanical Information and Ecology Network (BIEN) database. Methods Ecol. Evol. 9, 373–379 (2017).

61. Rautio, P., Fürst, A., Stefan, K., Raitio, H. & Bartels, U. Sampling and analysis of needles and leaves (ICP Forests, 2010).

62. Cayuela, L. & Oksanen, J. Taxonstand: Taxonomic Standardization of Plant Species Names (2016).

63. Qian, H. & Jin, Y. An updated megaphylogeny of plants, a tool for generating plant phylogenies and an analysis of phylogenetic community structure. J. Plant Ecol. 9, 233–239 (2016).

64. R Development Core Team R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, 2018).

65. Revell, L. J. phytools: an R package for phylogenetic comparative biology (and other things). Methods Ecol. Evol. 3, 217–223 (2012).

66. Tung Ho, L. S. & Ané, C. A linear-time algorithm for Gaussian and non-Gaussian trait evolution models. Syst. Biol. 63, 397–408 (2014).

67. Breheny, P. & Burchett, W. Visualization of regression models using visreg. R. J. 9:2, 56–71 (2017).

68. Bollback, J. P. SIMMAP: stochastic character mapping of discrete traits on phylogenies. BMC Bioinformatics 7, 88 (2006).

69. O’Meara, B. C., Ané, C., Sanderson, M. J. & Wainwright, P. C. Testing for different rates of continuous trait evolution using likelihood. Evolution 60, 922–933 (2006).

70. Thomas, G. H., Freckleton, R. P. & Székely, T. Comparative analyses of the influence of developmental mode on phenotypic diversification rates in shorebirds. Proc. R. Soc. Lond. B 273, 1619–1624 (2006).

71. Beaulieu, J. M., Jhwueng, D. C., Boettiger, C. & O’Meara, B. C. Modeling stabilizing selection: expanding the Ornstein–Uhlenbeck model of adaptive evolution. Evolution 66, 2369–2383 (2012).

72. Beaulieu, J. M. & O’Meara, B. OUwie: Analysis of Evolutionary Rates in an OU Framework (2016).

AcknowledgementsThis research was supported by the Spanish Government project CGL2016-79835-P (FERTWARM), European Research Council Synergy Grant ERC-2013-726 SyG-610028 IMBALANCE-P and Catalan Government project SGR 2017-1005. M.F.-M. is a postdoctoral fellow of the Research Foundation – Flanders (FWO). M.B. was supported by (Polish) NSF grants Sonatina 2017/24/C/NZ8/00151 and Uwertura 2018/28/U/NZ8/00003. This research was also supported by NSF grants DEB-1745496 630 to J.M.L. and DEB-1256394 to W.D.K.

Author contributionsM.F.-M., I.P. and I.A.J. conceived the paper. M.F.-M and F.S. analysed the data. M.F.-M., J.S., J.P., I.P., W.D.K. and J.M.L. provided the data. All authors, including M.B., A.C., A.H.-P., G.V. and J.M.E. contributed substantially to the writing and discussion of the paper.

Competing interestsThe authors declare no competing interests.

Additional informationExtended data is available for this paper at https://doi.org/10.1038/s41477-019-0549-y.

Supplementary information is available for this paper at https://doi.org/10.1038/s41477-019-0549-y.

Correspondence and requests for materials should be addressed to M.F.-M.

Peer review information Nature Plants thanks Shuli Niu, Ignacio Perez Ramos and the other, anonymous, reviewer for their contribution to the peer review of this work.

Reprints and permissions information is available at www.nature.com/reprints.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

© The Author(s), under exclusive licence to Springer Nature Limited 2019

NAturE PLANtS | VOL 5 | DECEMBER 2019 | 1222–1228 | www.nature.com/natureplants1228

ArticlesNaTUrE PlaNTs

Extended Data Fig. 1 | Evolutionary relationship between potential resource depletion coefficient (Ar1) and temporal variability (PV) in seed production. Evolutionary relationship between potential resource depletion coefficient (AR1) and temporal variability (PV) in seed production shown in a continuous trait phylogenetic reconstruction (a) and a phylomorphospace plot (b). Phylogenetic signal was estimated using Pagel’s lambda (λ). Potential resource depletion and variability in seed production were not evolutionary correlated. Negative values of AR1 indicate that potential resource depletion may happen, see Methods. N=219 species. t- value of the Pearson’s correlation was 1.95 (218 DF).

NAturE PLANtS | www.nature.com/natureplants

Articles NaTUrE PlaNTs

Extended Data Fig. 2 | Mean differences (ΔAICc, second-order Akaike information criterion) between each of the model’s AICc and the model with the lowest AICc. Mean differences (ΔAICc, second-order Akaike information criterion) between each of the model’s AICc and the model with the lowest AICc. Evolutionary models were Brownian motion (BM1, BMS) and generalised Ornstein-Uhlenbeck-based Hansen (OU1, OUM, OUMV), fitting “masting” and “non-masting” species-state and foliar nutrient concentrations (N: nitrogen, P: phosphorus, N:P: ratio N-to-P and, N×P: N times P (overall nutrient availability). Average AICc values were calculated using the subset of models in which none of them presented negative eigenvalues (sound models, n column: samples, independent simulations). Non-masting and masting columns indicate the number of species used in each category depending on the percentile of masting intensity used to classify species as non-masting (that is, higher than for example, 33%) and masting (that is, lower than for example, 66%). Models with ΔAICc lower than 2 (indicating equal performance) were highlighted. See Methods for further information.

NAturE PLANtS | www.nature.com/natureplants

ArticlesNaTUrE PlaNTs

Extended Data Fig. 3 | Phylogenetic tree including the subset of low (non-masting) and high masting intensity (masting) species used to perform the generalised Ornstein-uhlenbeck model results. Phylogenetic tree including the subset of low (non- masting) and high masting intensity (masting) species used to perform the generalised Ornstein-Uhlenbeck model results presented in the main text (20th – 80th percentile thresholds for non-masting and masting species, Fig. 3, Extended Data Fig. 2 and Extended Data Fig. 4). The phylogenetic tree includes the estimated probability that ancestor nodes were masting or non-masting species (large circles) as pie charts. Small circles indicate the current category of the species. The ancestral character reconstruction was performed using 1000 stochastic character-mapped trees (see Methods for further information).

NAturE PLANtS | www.nature.com/natureplants

Articles NaTUrE PlaNTs

Extended Data Fig. 4 | Estimated foliar nitrogen (N) and phosphorus (P) concentrations, N:P and N×P (overall nutrient availability) optimal values for masting and non-masting species. Estimated foliar nitrogen (N) and phosphorus (P) concentrations, N:P and N×P (overall nutrient availability) optimal values for masting and non-masting species using OUMV and OUM models (see Methods for further information about the models), chosen based on the lowest ΔAICc estimating different state means for masting and non-masting species (Extended Data Fig. 2). Masting and non-masting species were classified depending on the percentile of masting intensity (for example, masting for higher than 66%, non-masting for lower than 33%, see subheaders within the table). Columns 2.5%, 50 and 97.5% indicate, for masting and non-masting species, the percentiles of the optimal values based on the sound models (without negative eigenvalues, n column: samples, independent simulations) used. M>N% indicate the percentage of models in which masting species presented average higher N, P, N:P or N×P optimal values than non-masting species. ΔM-N, followed by s.e.m (standard error of the mean), indicate the paired (across simulations) difference between optimal values in masting and non-masting species. P (two-sided t-test) shows the P-value of the paired t-test testing for differences in the mean optimal values of masting and non-masting species. ΔM-N%, followed by s.e.m., indicates the average percentual difference (geometric, paired differences) in mean optimal values between masting and non-masting species.

NAturE PLANtS | www.nature.com/natureplants

ArticlesNaTUrE PlaNTs

Extended Data Fig. 5 | Evolutionary relationship between foliar N and P shown in a continuous trait phylogenetic reconstruction (a) and a phylomorphospace plot (b). Evolutionary relationship between foliar N and P shown in a continuous trait phylogenetic reconstruction (a) and a phylomorphospace plot (b). Phylogenetic signal was estimated using Pagel’s lambda (λ). Foliar N and P concentrations were evolutionary correlated. N=168 species. t-value of the Pearson’s correlation was 5.38 (166 DF).

NAturE PLANtS | www.nature.com/natureplants

Articles NaTUrE PlaNTs

Extended Data Fig. 6 | Map showing interannual variability (PV index) in mean annual precipitation (MAP) and site of origin of our fruit production data (blue dots). Map showing interannual variability (PV index) in mean annual precipitation (MAP) and site of origin of our fruit production data (blue dots).

NAturE PLANtS | www.nature.com/natureplants

1

nature research | reporting summ

aryO

ctober 2018

Corresponding author(s): Marcos Fernández-Martínez

Last updated by author(s): 18/9/19

Reporting SummaryNature Research wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Research policies, see Authors & Referees and the Editorial Policy Checklist.

StatisticsFor all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.

n/a Confirmed

The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement

A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly

The statistical test(s) used AND whether they are one- or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.

A description of all covariates tested

A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons

A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals)

For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted Give P values as exact values whenever suitable.

For Bayesian analysis, information on the choice of priors and Markov chain Monte Carlo settings

For hierarchical and complex designs, identification of the appropriate level for tests and full reporting of outcomes

Estimates of effect sizes (e.g. Cohen's d, Pearson's r), indicating how they were calculated

Our web collection on statistics for biologists contains articles on many of the points above.

Software and codePolicy information about availability of computer code

Data collection Data was collected through literature review and previously published databases (e.g., TRY, ICP Forests). The origin of all the data used is stated in the Methods section of the paper.

Data analysis We used R statistical software (v. 3.5.1) to analyse our data. All statistical methods and packages used are stated in the Methods.

For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors/reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Research guidelines for submitting code & software for further information.

DataPolicy information about availability of data

All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: - Accession codes, unique identifiers, or web links for publicly available datasets - A list of figures that have associated raw data - A description of any restrictions on data availability

The authors declare that the data supporting the findings of this study will be made openly available upon acceptance at USGS ScienceBase.

Field-specific reportingPlease select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.

Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences

2

nature research | reporting summ

aryO

ctober 2018

For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf

Ecological, evolutionary & environmental sciences study designAll studies must disclose on these points even when the disclosure is negative.

Study description We tested whether foliar nutrient concentrations was related to masting behaviour using a global dataset containing foliar nutrient concentrations and seed production for 219 plant species.

Research sample Our research sample were 219 plant species distributed across the globe. This data was previously used in a publication: Pearse et al., 2017. We also used foliar concentrations of N and P coming from the TRY database, the ICP forests, the Catalan Forest Inventory and amongst other smaller databases and from literature reviews carried out by coauthors.

Sampling strategy This is paper uses data collected from published literature on seed and fruit production and the elemental composition of leaves (from databases such as TRY and BIEN). Data on interannual reproductive effort (seed or fruit production) were compiled from Web of Science searches, scanning the literature cited of published papers to look for more records of reproductive effort, contacting managers of wildlife surveys, forestry districts, and regional seed surveys, and soliciting datasets in the Ecolog listserv (https://listserv.umd.edu/archives/ecolog-l.html). See Methods for more information. We collected all available data.

Data collection Data was collected from pre-exisiting databases (TRY, BIEN) and from a literature review (explained above). Compilation of seed and fruit production was carried on by Drs. Pearse, LaMontagne and Koenig. They cleaned the database to include only records that matched certain characteristics (see Data exclusions). Drs. Sardans, Peñuelas, Fernández-Martínez collected foliar nutrient concentrations (from TRY and BIEN databases). TRY and BIEN databases are already curated. We downloaded the complete databases and merged it with the species for which we had fruit production data.

Timing and spatial scale Records of seed production were irregularly comprised between the years 1900-2014. The spatial scale of this dataset is global (see figure ED6).

Data exclusions We pre-established several criteria to include records in our dataset. We only included records when: i) data was available for more than four consecutive years per species at a given site (with clear coordinates), ii) records could be clearly assigned to plants of a particular species, iii) records were not measured in such an indirect way their accuracy could be jeopardised (e.g., anticipated correlations with game abundance), iv) data represented seed or fruit production, or inflorescence set only for those cases where inflorescences are strongly linked to seed or fruit production, and v) records from iteroparous perennial plants whose seed set could not be explained by changes in population size.

Reproducibility No experiments were performed. Reproducibility is ensured by providing the collected dataset, for which anyone will be able to check values of seed production and average plant foliar N and P concentrations.

Randomization We did not perform any experiment, therefore, randomization was not necessary.

Blinding We did not run any experiment, so no blinding was performed.

Did the study involve field work? Yes No

Reporting for specific materials, systems and methodsWe require information from authors about some types of materials, experimental systems and methods used in many studies. Here, indicate whether each material, system or method listed is relevant to your study. If you are not sure if a list item applies to your research, read the appropriate section before selecting a response.

Materials & experimental systemsn/a Involved in the study

Antibodies

Eukaryotic cell lines

Palaeontology

Animals and other organisms

Human research participants

Clinical data

Methodsn/a Involved in the study

ChIP-seq

Flow cytometry

MRI-based neuroimaging